openbsd
5 years agochange valid_domainname() to accept a trailing dot.
eric [Wed, 28 Aug 2019 19:46:20 +0000 (19:46 +0000)]
change valid_domainname() to accept a trailing dot.

ok millert@

5 years agoThis code calls task_add(9) from a timeout, so create the taskq with
kettenis [Wed, 28 Aug 2019 19:37:56 +0000 (19:37 +0000)]
This code calls task_add(9) from a timeout, so create the taskq with
IPL_SOFTCLOCK instead of IPL_NONE.

ok mpi@, deraadt@, patrick@

5 years agoAdd a link-greeting report. This allows us to get the active domain name
martijn [Wed, 28 Aug 2019 15:50:36 +0000 (15:50 +0000)]
Add a link-greeting report. This allows us to get the active domain name
in use for the current request.

OK gilles@

5 years agosync
deraadt [Wed, 28 Aug 2019 15:49:30 +0000 (15:49 +0000)]
sync

5 years agoMake filters more developer-friendly by giving more feedback on why a
martijn [Wed, 28 Aug 2019 15:37:28 +0000 (15:37 +0000)]
Make filters more developer-friendly by giving more feedback on why a
message was rejected.

Feedback semarie@
OK gilles@

5 years agoIntroduce SCSI0(), SCSI2() and SCSI3() defines to clarify (some) uses
krw [Wed, 28 Aug 2019 15:17:23 +0000 (15:17 +0000)]
Introduce SCSI0(), SCSI2() and SCSI3() defines to clarify (some) uses
of SCSISPC() when checking the values of the INQUIRY version field.

5 years agosys/arch/luna88k/luna88k/locore0.S:dummy_cpu assumes that
aoyama [Wed, 28 Aug 2019 13:48:40 +0000 (13:48 +0000)]
sys/arch/luna88k/luna88k/locore0.S:dummy_cpu assumes that
ci_mp_atomic_{begin,end} are 6th and 7th elements of cpu_info
structure.  Actually that is dummy structure used in early boot stage,
but for consistency, move ci_srp_hazards position in cpu_info.

ok mpi@

5 years agoadd config lines for amdgpu on arm64
jsg [Wed, 28 Aug 2019 10:45:25 +0000 (10:45 +0000)]
add config lines for amdgpu on arm64

5 years agonew manual page AES_encrypt(3)
schwarze [Wed, 28 Aug 2019 10:37:42 +0000 (10:37 +0000)]
new manual page AES_encrypt(3)

5 years agoBuild and enable amdgpu(4) on arm64. The DCN1.0 support has been made
kettenis [Wed, 28 Aug 2019 10:17:59 +0000 (10:17 +0000)]
Build and enable amdgpu(4) on arm64.  The DCN1.0 support has been made
optional and will only be compiled for amd64/i386.  Apparently this is
only needed on Raven Ridge APUs.

ok jsg@, patrick@, deraadt@

5 years agoamlpinctrl(4)
kettenis [Wed, 28 Aug 2019 09:46:32 +0000 (09:46 +0000)]
amlpinctrl(4)

5 years agoThe resize event was never deciding to actually resize the pane if there
nicm [Wed, 28 Aug 2019 07:34:32 +0000 (07:34 +0000)]
The resize event was never deciding to actually resize the pane if there
was output in the pane faster than the timer would fire, so change how
it works to only defer the timer again if the pane was actually resized
within the last timer period. Reported by James Tai in GitHub issue
1880.

5 years agoAdd amlpinctrl(4), a driver to configure pins and control GPIOs on Amlogic
kettenis [Wed, 28 Aug 2019 07:12:37 +0000 (07:12 +0000)]
Add amlpinctrl(4), a driver to configure pins and control GPIOs on Amlogic
SoCs.  For now only a limited set of GPIO pins on the G12A/G12B variants
are supported.

ok patrick@

5 years agoAdd a PHY "framework".
kettenis [Wed, 28 Aug 2019 07:03:51 +0000 (07:03 +0000)]
Add a PHY "framework".

ok patrick@

5 years agotypo
naddy [Wed, 28 Aug 2019 06:11:36 +0000 (06:11 +0000)]
typo

5 years agoimplement a DV_POWERDOWN handler to flush cache and shutdown the controller
dlg [Wed, 28 Aug 2019 04:55:51 +0000 (04:55 +0000)]
implement a DV_POWERDOWN handler to flush cache and shutdown the controller

this has been in snaps for the last week without issue, and has
been running in production on a bunch of my boxes for a week before
that, also without issue.

5 years agosync
deraadt [Wed, 28 Aug 2019 03:56:11 +0000 (03:56 +0000)]
sync

5 years agos/No separable connector/No connector/
dlg [Wed, 28 Aug 2019 01:52:38 +0000 (01:52 +0000)]
s/No separable connector/No connector/

shrinks some QSFP+ DAC output a bit.

5 years agoremove some cruft
dlg [Wed, 28 Aug 2019 01:50:48 +0000 (01:50 +0000)]
remove some cruft

5 years agotry to factor media information out and use it for qsfp as well as sfp.
dlg [Wed, 28 Aug 2019 01:47:20 +0000 (01:47 +0000)]
try to factor media information out and use it for qsfp as well as sfp.

qsfp and xfp have remarkably similar layouts, so we can use the
same code to print media for both. sfp and xfp/qsfp generally have
the same fields, just in different register locations and with some
different scales/factors for some values. this change provides a
map of these differences as structs for the sfp and qsfp/xfp layouts.

the big difference between sfp, xfp, and qsfp that still remains
is how they interpret the wavelength field. qsfp stores either optic
wavelength in units of 0.05nm, or a couple of copper cable attenuation
values in dB at different frequencies. sfp stores wavelength in nm
units, but has magic values to blacklist copper cables with. xfp
stores wavelength in 0.05 nm units, so more like like qsfp.  right
now the code uses the sfp behaviour, which means qsfp will show a
weird wavelength for copper cables when it shouldnt. i'm still
trying to figure out the least worst way to handle that, but at
least it looks right for optics now.

5 years agoprint more qsfp temperature information
dlg [Tue, 27 Aug 2019 23:24:35 +0000 (23:24 +0000)]
print more qsfp temperature information

this shows the maximum case temperature reported by the chip, and
the low and high alarm ranges for temp and vcc when page 3 is
available and working.

this will hopefully give deraadt@ a better idea about how much
headroom one of his 100g optics has.

5 years agoOn amd64 the bootup and resume paths for SP/BP/SP were not performing
deraadt [Tue, 27 Aug 2019 22:39:51 +0000 (22:39 +0000)]
On amd64 the bootup and resume paths for SP/BP/SP were not performing
msr/pat, clock-startup, firmware upgrade, fpu initialization, and cpu
soft-state initialization in the same order.  That kind of chaos is
crazy and fragile. Improve the acpi resume call sequence and unify order
of the operations.

5 years agoamlrng(4)
kettenis [Tue, 27 Aug 2019 22:27:08 +0000 (22:27 +0000)]
amlrng(4)

5 years agoAdd amlrng(4), the simples random numer generator driver ever.
kettenis [Tue, 27 Aug 2019 22:21:52 +0000 (22:21 +0000)]
Add amlrng(4), the simples random numer generator driver ever.

5 years agoHonour -n when deleting entries, sync parse_host() from ndp(8)
kn [Tue, 27 Aug 2019 20:50:36 +0000 (20:50 +0000)]
Honour -n when deleting entries, sync parse_host() from ndp(8)

arp(8) already has a common getinetaddr() but it ignores -n for delete() and
uses gethostbyname(3), so simply replace it with parse_host() from ndp.c
instead of changing behaviour around the old API.

In set(), make host[] and eaddr[] const since all functions receiving those
variables as parameters expect const anyway.

OK bluhm

5 years agoSync const correctness with arp(8)
kn [Tue, 27 Aug 2019 20:42:40 +0000 (20:42 +0000)]
Sync const correctness with arp(8)

Zap obsolete linter comments while here.

No object change;
OK bluhm

5 years agoSync example with current databases/postgresql flags
awolk [Tue, 27 Aug 2019 20:05:02 +0000 (20:05 +0000)]
Sync example with current databases/postgresql flags

OK sthen@, pea@

5 years agoFix HSET4 and HCLR4 macros; I shouldn't make last-minute changes without
kettenis [Tue, 27 Aug 2019 18:29:45 +0000 (18:29 +0000)]
Fix HSET4 and HCLR4 macros; I shouldn't make last-minute changes without
actually testing them.

5 years agosync
jcs [Tue, 27 Aug 2019 18:00:19 +0000 (18:00 +0000)]
sync

5 years agoadd ethernet found on Lenovo USB-C Dock
jcs [Tue, 27 Aug 2019 17:59:58 +0000 (17:59 +0000)]
add ethernet found on Lenovo USB-C Dock

5 years agosync
deraadt [Tue, 27 Aug 2019 17:03:02 +0000 (17:03 +0000)]
sync

5 years agosync
sthen [Tue, 27 Aug 2019 16:58:38 +0000 (16:58 +0000)]
sync

5 years agodisable base-gcc on i386, from daniel@, ok deraadt
sthen [Tue, 27 Aug 2019 15:32:33 +0000 (15:32 +0000)]
disable base-gcc on i386, from daniel@, ok deraadt

5 years agoKeep ieee80211_node structures cached across scans, rather than
stsp [Tue, 27 Aug 2019 14:57:48 +0000 (14:57 +0000)]
Keep ieee80211_node structures cached across scans, rather than
wiping the entire cache every time a scan is triggered.

This has benefits for userland programs trying to look at scan results,
and for drivers which don't get a full view from hardware on every scan.

Nodes will still be evicted from the cache in one of several ways:

Add a new way of timing out inactive nodes which don't send a beacon
within 10 scan iterations, for client mode only. This should get rid
of stale APs if we're scanning for some time in a changing environment.

If we fail to associate a few times, the corresponding node is removed.

If net80211 transitions back to INIT state (e.g. because of a
user-initiated configuration change) all nodes are removed.

When a background scan starts all nodes will be removed. This could be
revisited later but has been intentionally left as-is for now.

Tested by myself, jmc@, benno@, procter@ on iwm(4), athn(4) and ral(4).

5 years agoRefactor probing logic to mirror detach logic. i.e. put smarts in
krw [Tue, 27 Aug 2019 14:41:45 +0000 (14:41 +0000)]
Refactor probing logic to mirror detach logic. i.e. put smarts in
scsi_probe() and make scsi_probe_bus(), scsi_probe_target() and
scsi_probe_lun() simple wrappers around scsi_probe().

Abstract the determination of which luns to probe into a separate
function. Thus eliminating the need to remove/add lun 0 link while
probing devices modern enough to support REPORTLUNS. Which means the
lun 0 link is no longer in different positions in the scsi_link list
for such devices compared to older devices which are blindly probed
until an invalid LUN is encountered.

5 years agog77 is not a valid MODFORTRAN_COMPILER any more
jca [Tue, 27 Aug 2019 14:33:24 +0000 (14:33 +0000)]
g77 is not a valid MODFORTRAN_COMPILER any more

5 years agoAdd imxtmu(4), a driver to support the temperature sensors on
patrick [Tue, 27 Aug 2019 12:51:57 +0000 (12:51 +0000)]
Add imxtmu(4), a driver to support the temperature sensors on
the i.MX8M SoCs.

ok kettenis@

5 years agohave a go at printing more monitoring information.
dlg [Tue, 27 Aug 2019 11:54:42 +0000 (11:54 +0000)]
have a go at printing more monitoring information.

this is a bit rough, but a good start. apart from the media types
and length, qsfp support should be on par with sfp modules now.

5 years agoError paths in radeondrm(4) call unregister_acpi_notifier() unconditionally
kettenis [Tue, 27 Aug 2019 11:46:07 +0000 (11:46 +0000)]
Error paths in radeondrm(4) call unregister_acpi_notifier() unconditionally
so check whether a notifier is on the list before trying to remove it.
Fixes a crash found by semarie@.

ok semarie@

5 years agodocument OCSP_parse_url(3)
schwarze [Tue, 27 Aug 2019 10:48:41 +0000 (10:48 +0000)]
document OCSP_parse_url(3)

5 years agodocument OCSP_cert_status_str(3)
schwarze [Tue, 27 Aug 2019 10:00:41 +0000 (10:00 +0000)]
document OCSP_cert_status_str(3)

5 years agodocument OCSP_response_status_str(3)
schwarze [Tue, 27 Aug 2019 09:40:29 +0000 (09:40 +0000)]
document OCSP_response_status_str(3)

5 years agoamlclock(4)
kettenis [Tue, 27 Aug 2019 09:05:57 +0000 (09:05 +0000)]
amlclock(4)

5 years agoamlclock(4)
kettenis [Tue, 27 Aug 2019 08:24:37 +0000 (08:24 +0000)]
amlclock(4)

5 years agoamclock(4), a driver for the Amlogic SoC clocks. For now this implements
kettenis [Tue, 27 Aug 2019 08:19:55 +0000 (08:19 +0000)]
amclock(4), a driver for the Amlogic SoC clocks.  For now this implements
a tiny number of clocks for the G12A/B families only.

ok jsg@

5 years agoBetter error reporting on malformed packets.
martijn [Tue, 27 Aug 2019 06:14:28 +0000 (06:14 +0000)]
Better error reporting on malformed packets.

5 years agodon't check the page number was set correctly.
dlg [Tue, 27 Aug 2019 04:26:50 +0000 (04:26 +0000)]
don't check the page number was set correctly.

there are some extremely terrible modules out there that violate
specs badly. some SFP+ DACs i have don't support page switching,
and always report page 0xff. let userland look at them anyway.

5 years agouse m_getptr to get to the right mbuf and offset for the ttl in mpls_gettl.
dlg [Tue, 27 Aug 2019 00:49:48 +0000 (00:49 +0000)]
use m_getptr to get to the right mbuf and offset for the ttl in mpls_gettl.

problem found by and this fix was tested by groos at xiplink dot com
on bugs@

5 years agomake SIOCGIFSFFPAGE support QSFP modules better.
dlg [Tue, 27 Aug 2019 00:40:26 +0000 (00:40 +0000)]
make SIOCGIFSFFPAGE support QSFP modules better.

basically the "read phy register" admin op uses the "address" field
on SFP modules for the i2c address, but uses that same field for
the "sff page number" for QSFP. this wraps the calls to the admin
op with sfp and qsfp specific code, and chooses between them depending
on what type of phy the fw reports.

tested by me and jmatthew@ pretty hard

5 years agomake a start at reporting qsfp monitor values.
dlg [Tue, 27 Aug 2019 00:33:57 +0000 (00:33 +0000)]
make a start at reporting qsfp monitor values.

i have some qsfp DACs and a couple of optics, and they're all
terrible, so this is about as far as i can go for now. at least the
code will be robust in the face of terrible modules though.

the DACs are pretty dumb and basically report that they're DACs
with some strings. this code just prints that they're DACs with
strings now.

modules are supposed to be able to report overall temperature and
voltage, and optics can report tx and rx values for the 4 different
signal lanes they're supposed to provide. interestingly the current
values are always reported in the lower page, but thresholds are
reported in page 3, but not all modules support page switching.
devices are supposed to say whether they can switch pages, but i
have one that does say it can switch but then doesn't. anyway, the
take away is that it is therefore possible for a module to report
values without also
reporting thresholds.

this sets the code up to report the values on their own if we can't
query page 3 for any reason.

if the temp sensor value looks bogus (ie, 0x0000 or 0xffff), assume
the monitor values are bogus and bail early.

hopefully i can find a module soon that supports multiple signal
lanes and actually reports their values and thresholds for them.

5 years agoreplace a mmap() with pread(). The mmap() is not needed, the file is
benno [Mon, 26 Aug 2019 22:22:14 +0000 (22:22 +0000)]
replace a mmap() with pread(). The mmap() is not needed, the file is
read sequentially anyway, (hash-sized) block by block. This avoids
counting potentially large files against the memory limits of the
process. Problem reported by Jan Stary, thanks.
ok deraadt@

5 years agosync
deraadt [Mon, 26 Aug 2019 18:56:30 +0000 (18:56 +0000)]
sync

5 years agoWhen a thread tries to exclusively lock a vnode, the same thread must
anton [Mon, 26 Aug 2019 18:56:29 +0000 (18:56 +0000)]
When a thread tries to exclusively lock a vnode, the same thread must
ensure that any other thread currently trying to acquire the underlying
vnode lock has observed that the same vnode is about to be exclusively
locked. Such threads must then sleep until the exclusive lock has been
released and then try to acquire the lock again. Otherwise, exclusive
access to the vnode cannot be guaranteed.

Thanks to naddy@ and visa@ for testing; ok visa@

Reported-by: syzbot+374d0e7e2400004957f7@syzkaller.appspotmail.com
5 years agoFix file descriptor leak due to popfile() never closing the main config file.
tobhe [Mon, 26 Aug 2019 18:53:58 +0000 (18:53 +0000)]
Fix file descriptor leak due to popfile() never closing the main config file.

The fix is the same as for other parse.y files in the tree (see bgpd(8) or
unwind(8))

ok bluhm@

5 years agofree(3) style functions should accept NULL and do nothing
pamela [Mon, 26 Aug 2019 18:50:04 +0000 (18:50 +0000)]
free(3) style functions should accept NULL and do nothing
OK florian

5 years agoDo not use the flow of the first fragment to store ECN information.
bluhm [Mon, 26 Aug 2019 18:47:53 +0000 (18:47 +0000)]
Do not use the flow of the first fragment to store ECN information.
Handle the ECN in the fragment queue.
Reported-by: syzbot+0aa80b25e9041001cac8@syzkaller.appspotmail.com
fix from FreeBSD; OK claudio@

5 years agoFix file descriptor leak in config parser. Inspired by bgpd parse.y.
tobhe [Mon, 26 Aug 2019 16:41:08 +0000 (16:41 +0000)]
Fix file descriptor leak in config parser. Inspired by bgpd parse.y.

ok patrick@

5 years agoFix swap-window -d to work as intended, GitHub issue 1879 from Sam Stuewe.
nicm [Mon, 26 Aug 2019 16:35:41 +0000 (16:35 +0000)]
Fix swap-window -d to work as intended, GitHub issue 1879 from Sam Stuewe.

5 years agoAdjust umb DNS printing code since the type of those addresses changed.
claudio [Mon, 26 Aug 2019 15:23:49 +0000 (15:23 +0000)]
Adjust umb DNS printing code since the type of those addresses changed.
OK gerhard@

5 years agoReplace umb_ntop() with sockaddr_ntop() which does almost the same thing.
claudio [Mon, 26 Aug 2019 15:23:01 +0000 (15:23 +0000)]
Replace umb_ntop() with sockaddr_ntop() which does almost the same thing.
Also change the storage type for the DNS addresses to struct in_addr since
that is more convinient for userland. This includes some minor other cleanup.
OK gerhard@

5 years agodocument i2a_ASN1_INTEGER(3)
schwarze [Mon, 26 Aug 2019 12:45:27 +0000 (12:45 +0000)]
document i2a_ASN1_INTEGER(3)

5 years agoRemove rdtsc macro.
pirofti [Mon, 26 Aug 2019 12:41:47 +0000 (12:41 +0000)]
Remove rdtsc macro.

This is legacy code and was probably used instead of the desried
inline'd function in cpufunc.h.

OK deraadt@, kettenis@.

5 years agodocument ASN1_put_object(3) and ASN1_put_eoc(3)
schwarze [Mon, 26 Aug 2019 11:41:31 +0000 (11:41 +0000)]
document ASN1_put_object(3) and ASN1_put_eoc(3)

5 years agoRemove unnecessary retrieval of the GPC regmap. We don't need it since
patrick [Mon, 26 Aug 2019 11:24:03 +0000 (11:24 +0000)]
Remove unnecessary retrieval of the GPC regmap.  We don't need it since
the imxgpc(4)'s power domain framework will take care of the registers
and thus dwpcie(4) hasn't touched the registers itself anyway.

ok kettenis@

5 years agoremove some debug cruft that i should have removed after the addr64 stuff
dlg [Mon, 26 Aug 2019 11:05:54 +0000 (11:05 +0000)]
remove some debug cruft that i should have removed after the addr64 stuff

5 years agoremove redundant assignment to ic_curmode
stsp [Mon, 26 Aug 2019 09:53:31 +0000 (09:53 +0000)]
remove redundant assignment to ic_curmode

This assignment to ic_curmode is redundant because it already occurs
inside ieee80211_setmode(), and channel information in selbs and ni
is equivalent after node_copy().

ok mpi@ kevlo@

5 years agoamlreset(4)
kettenis [Mon, 26 Aug 2019 09:44:26 +0000 (09:44 +0000)]
amlreset(4)

5 years agoAdd amlreset(4), a driver to support reset signal controller blocks on
kettenis [Mon, 26 Aug 2019 09:42:31 +0000 (09:42 +0000)]
Add amlreset(4), a driver to support reset signal controller blocks on
Amlogic SoCs.

ok patrick@

5 years agoImplement GPIO hogging.
kettenis [Mon, 26 Aug 2019 09:22:27 +0000 (09:22 +0000)]
Implement GPIO hogging.

ok patrick@

5 years agopf.conf "set timeout interval 1" causes kernel crash
sashan [Mon, 26 Aug 2019 09:19:12 +0000 (09:19 +0000)]
pf.conf "set timeout interval 1" causes kernel crash
(bug reported and fix tested by Kor)

ok kn@

5 years agoamluart(4)
kettenis [Mon, 26 Aug 2019 09:16:39 +0000 (09:16 +0000)]
amluart(4)

5 years agoAdd amluart(4), a driver for the UARTs found on various Amlogic SoCs.
kettenis [Mon, 26 Aug 2019 09:10:22 +0000 (09:10 +0000)]
Add amluart(4), a driver for the UARTs found on various Amlogic SoCs.

ok jsg@

5 years agodocument ASN1_OCTET_STRING_cmp(3), ASN1_OCTET_STRING_dup(3), and
schwarze [Mon, 26 Aug 2019 07:59:02 +0000 (07:59 +0000)]
document ASN1_OCTET_STRING_cmp(3), ASN1_OCTET_STRING_dup(3), and
ASN1_OCTET_STRING_set(3)

5 years agoalso pass the signature style
espie [Mon, 26 Aug 2019 06:52:51 +0000 (06:52 +0000)]
also pass the signature style

5 years agosync
deraadt [Sun, 25 Aug 2019 20:39:46 +0000 (20:39 +0000)]
sync

5 years agoChange generating and checking of primes so that the error rate of
schwarze [Sun, 25 Aug 2019 19:23:59 +0000 (19:23 +0000)]
Change generating and checking of primes so that the error rate of
not being prime depends on the intended use based on the size of
the input.  For larger primes this will result in more rounds of
Miller-Rabin.  The maximal error rate for primes with more than
1080 bits is lowered to 2^-128.
Patch from Kurt Roeckx <kurt@roeckx.be> and Annie Yousar
via OpenSSL commit feac7a1c Jul 25 18:55:16 2018 +0200,
still under a free license.
OK tb@.

5 years agosynch with other changes. bug noticed by naddy@
espie [Sun, 25 Aug 2019 19:06:36 +0000 (19:06 +0000)]
synch with other changes. bug noticed by naddy@

5 years agoAdd SIOCDIFADDR_IN6 to the wroute pledge to allow removal of IPv6 addresses
pamela [Sun, 25 Aug 2019 18:46:40 +0000 (18:46 +0000)]
Add SIOCDIFADDR_IN6 to the wroute pledge to allow removal of IPv6 addresses
Needed for future slaacd(8) improvements

OK florian@ deraadt@

5 years agodocument EVP_sm3(3) and EVP_whirlpool(3), loosely based on the
schwarze [Sun, 25 Aug 2019 17:08:20 +0000 (17:08 +0000)]
document EVP_sm3(3) and EVP_whirlpool(3), loosely based on the
OpenSSL 1.1.1 pages, which are still under a free license

5 years agoMerge common parsing route into new parse_host() helper
kn [Sun, 25 Aug 2019 16:23:36 +0000 (16:23 +0000)]
Merge common parsing route into new parse_host() helper

This chops 18 lines off while unifying the code, now that set(), get()
as well as delete() all parse hostnames and honour -n the same way.

OK deraadt

5 years agospace -> tabs
ajacoutot [Sun, 25 Aug 2019 15:50:21 +0000 (15:50 +0000)]
space -> tabs

ok deraadt@ kn@

5 years agofix reversed meaning of error codes;
schwarze [Sun, 25 Aug 2019 15:39:10 +0000 (15:39 +0000)]
fix reversed meaning of error codes;
from Martin Ukrop <mukrop at mail dot muni dot cz>
via OpenSSL commit bb00b040 Aug 5 14:14:54 2019 +0200

5 years agotypo in function argument type;
schwarze [Sun, 25 Aug 2019 15:17:19 +0000 (15:17 +0000)]
typo in function argument type;
from Jan Macku <jamacku at redhat dot com>
via OpenSSL commit a9b9d265 Jan 30 16:09:50 2019 +0100

5 years agoCorrectly document the return values of i2d_ECDSA_SIG(3) and
schwarze [Sun, 25 Aug 2019 14:11:41 +0000 (14:11 +0000)]
Correctly document the return values of i2d_ECDSA_SIG(3) and
d2i_ECDSA_SIG(3); triggered by OpenSSL commit da4ea0cf Aug 5 16:13:24
2019 +0100, but solved differently.  While here, adjust argument
placeholders and wording to our usual conventions, and don't try
to reiterate the complicated contents of ASN1_item_d2i(3) here.

5 years agodrm/amdgpu: fix a potential information leaking bug
jsg [Sun, 25 Aug 2019 12:02:59 +0000 (12:02 +0000)]
drm/amdgpu: fix a potential information leaking bug

From Wang Xiayang
68340db992f6704b86a27e3f9f91d44ce1c8cb31 in linux 4.19.y/4.19.68
929e571c04c285861e0bb049a396a2bdaea63282 in mainline linux

5 years agoavoid checking s->spf twice if it is above maximum in uaudio_adjspf(); ok ratchov@
miko [Sun, 25 Aug 2019 09:16:04 +0000 (09:16 +0000)]
avoid checking s->spf twice if it is above maximum in uaudio_adjspf(); ok ratchov@

5 years agoimport the CRYPTO_memcmp(3) manual from OpenSSL 1.1.1,
schwarze [Sun, 25 Aug 2019 06:20:22 +0000 (06:20 +0000)]
import the CRYPTO_memcmp(3) manual from OpenSSL 1.1.1,
still under a free license, tweaked by me

5 years agoAllow for "port smtp" and "port smtps" on listen statement.
martijn [Sun, 25 Aug 2019 03:40:45 +0000 (03:40 +0000)]
Allow for "port smtp" and "port smtps" on listen statement.
This was previously not allowed, because both smtp and smtps are keywords.
Since port's argument is non-optional and smtp and smtps are common enough
there is no chance for misinterpretation.

Problem reported by phatbuckett <at> gmail <dot> com
OK gilles@

5 years agoSimply logic of detaching things. scsi_detach_bus() folded into
krw [Sat, 24 Aug 2019 13:41:06 +0000 (13:41 +0000)]
Simply logic of detaching things. scsi_detach_bus() folded into
scsi_detach(), scsi_detach_target() and scsi_detach_lun() become
simple wrappers of scsi_detach() invocations.

No intentional functional change.

5 years agoClarify "protected-subnet" option.
tobhe [Sat, 24 Aug 2019 13:24:49 +0000 (13:24 +0000)]
Clarify "protected-subnet" option.

Explain the use of the option (according to the RFC) and make clear it is
not usually needed for subnets specified in "from" and "to" options.

ok sthen@

5 years agoFix conflict when IKE SA and Child SA rekeying happen at the same time.
tobhe [Sat, 24 Aug 2019 13:09:38 +0000 (13:09 +0000)]
Fix conflict when IKE SA and Child SA rekeying happen at the same time.

If the IKE SA changes during an ongoing rekey exchange, messages may be
lost because they were inteded for the old SA. An iked instance that is
waiting for a rekey Child SA response will no longer reply to IKE SA
rekey requests until the ongoing Child SA exchange has completed or
timed out.

ok sthen@

5 years agoKNF tweak.
krw [Sat, 24 Aug 2019 02:13:55 +0000 (02:13 +0000)]
KNF tweak.

5 years agosync devlist
jsg [Sat, 24 Aug 2019 01:45:59 +0000 (01:45 +0000)]
sync devlist

5 years agodrm/i915/cml: Introduce Comet Lake PCH
jsg [Sat, 24 Aug 2019 01:43:12 +0000 (01:43 +0000)]
drm/i915/cml: Introduce Comet Lake PCH

From Anusha Srivatsa
729ae330a0f2e270db2ca70c06a83d0aa2776288 in mainline linux

5 years agodrm/i915/cml: Add CML PCI IDS
jsg [Sat, 24 Aug 2019 01:39:55 +0000 (01:39 +0000)]
drm/i915/cml: Add CML PCI IDS

From Anusha Srivatsa
a7b4deeb02b978bc59808cb13c93ba84f01023a4 in mainline linux

5 years agoAssign the filter name to the proc name in the case of proc_exec.
martijn [Fri, 23 Aug 2019 19:05:01 +0000 (19:05 +0000)]
Assign the filter name to the proc name in the case of proc_exec.
This gives a more intuitive log-entry for filter stderr output and gets rid
of the last_dynproc_id.

"nice" gilles@

5 years agosync
deraadt [Fri, 23 Aug 2019 16:06:43 +0000 (16:06 +0000)]
sync

5 years agoHonour -n when setting entries
kn [Fri, 23 Aug 2019 15:41:59 +0000 (15:41 +0000)]
Honour -n when setting entries

Just like avoiding reverse lookups when printing entries or not resolving
hostnames when deleting entries, make -n effect setting entries as well;
just like pfctl(8)'s -N from which the manual wording was copied verbatim.

OK deraadt

5 years agoonly process records of the expected type.
eric [Fri, 23 Aug 2019 15:39:11 +0000 (15:39 +0000)]
only process records of the expected type.
fix an issue where CNAME records generate bogus results.

ok gilles@

5 years agodocument RSAPrivateKey_dup(3) and RSAPublicKey_dup(3)
schwarze [Fri, 23 Aug 2019 15:18:13 +0000 (15:18 +0000)]
document RSAPrivateKey_dup(3) and RSAPublicKey_dup(3)